2026 年第 10 期 第 21 卷

成人类鼻疽肺炎并发急性呼吸衰竭的预测模型构建及验证

Development and validation of a predictive model for acute respiratory failure complicated by melioidosis pneumonia in adult patients

作者：李少宁李慧媛周向东刘畅黄华萍李亮

英文作者：Li Shaoning Li Huiyuan Zhou Xiangdong Liu Chang Huang Huaping Li Liang

单位：海南医科大学第一附属医院呼吸与危重症医学科海南医科大学急救与创伤研究教育部重点实验室,海口570000

英文单位：Department of Respiratory and Critical Care Medicine the First Affiliated Hospital of Hainan Medical University Key Laboratory of Emergency and Trauma Research Ministry of Education Hainan Medical University Haikou 570000 China

关键词：类鼻疽肺炎；呼吸衰竭；预测模型

英文关键词：Melioidosispneumonia;Respiratoryfailure;Predictivemodel

• 摘要：

• 目的  分析成人类鼻疽肺炎患者并发急性呼吸衰竭（ARF）的危险因素，并构建列线图预测模型和验证其效能，提高对并发ARF的预警。方法 回顾性收集2005年2月至2025年7月就诊于海南医科大学第一附属医院的成人类鼻疽肺炎患者114例的临床资料。根据是否合并ARF分为不合并ARF组(60例)及合并ARF组(54例)。对比2组一般资料、临床症状、实验室检查等指标。单因素及多因素Logistic回归分析筛选成人类鼻疽肺炎并发ARF的危险因素，构建列线图模型、评估模型的预测能力。结果 2组发病时间、CURB-65评分分层及脑梗死比例比较差异均有统计学意义(均P＜0.05)。2组中性粒细胞百分比、中性粒细胞计数、淋巴细胞计数、中性粒细胞与淋巴细胞比值、血小板与淋巴细胞比值、淋巴细胞与血小板比值、血小板计数、纤维蛋白降解产物、D-二聚体、高敏C反应蛋白、乳酸脱氢酶（LDH）、白蛋白、总蛋白、球蛋白、丙氨酸转氨酶、天冬氨酸转氨酶、血尿素氮、血尿素氮与白蛋白比值、降钙素原、红细胞沉降率及病原微生物总数≥2种比例［25.9%（14/54）比10.0%（6/60）］、调整抗菌药物(根据培养结果)比例［81.5%（44/54）比33.3%（20/60）］比较差异均有统计学意义（均P＜0.05）。多因素分析提示CURB-65评分-中风险、中性粒细胞计数、LDH、丙氨酸转氨酶、病原微生物总数≥2种为成人类鼻疽肺炎并发ARF的独立危险因素(比值比=21.705、1.123、1.010、1.001、7.752，均P＜0.05)。基于Logistic回归筛选预测因子，构建预测成人类鼻疽肺炎患者并发ARF的列线图模型并验证。通过绘制受试者工作特征曲线评价模型区分度，模型曲线下面积为0.867(95%置信区间：0.700～1.000 )(P＜0.05)，以约登指数最大点为最佳临界点时，模型的敏感度为80.0%，特异度为90.0%。结论 本研究构建的模型对成人类鼻疽肺炎并发ARF具有良好的预测能力，可为临床早期预警、早期干预治疗提供参考依据。

• Objective To analyze the risk factors for acute respiratory failure (ARF) in adult patients with melioidosis pneumonia, construct a nomogram prediction model and verify its efficacy, so as to improve the early warning of complicated ARF. Methods Clinical data of 114 adult patients with melioidosis pneumonia admitted to the First Affiliated Hospital of Hainan Medical University from February 2005 to July 2025 were retrospectively collected. Patients were divided into non-ARF group (60 cases) and ARF group (54 cases) according to the presence or absence of ARF. General data, clinical symptoms, laboratory tests and other indicators were compared between the two groups. Univariate and multivariate Logistic regression analyses were performed to screen risk factors for ARF in adult patients with melioidosis pneumonia. A nomogram model was established and its predictive performance was evaluated. Results There were statistically significant differences in the disease onset time, CURB-65 score stratification and proportion of cerebral infarction between the two groups (all P＜0.05). There were statistically significant differences in neutrophil percentage, neutrophil count, lymphocyte count, neutrophil to lymphocyte ratio, platelet to lymphocyte ratio, lymphocyte to platelet ratio, platelet count, fibrin degradation product, D-dimer, high-sensitivity C-reactive protein, lactate dehydrogenase (LDH), albumin, total protein, globulin, alanine aminotransferase, aspartate aminotransferase, blood urea nitrogen, blood urea nitrogen-to-albumin ratio, procalcitonin, erythrocyte sedimentation rate, proportion of ≥2 pathogenic microorganisms ［25.9%(14/54) vs 10.0%(6/60)］ and proportion of adjusted antimicrobial agents (according to culture results)［81.5%(44/54) vs 33.3%(20/60)］ between the two groups (all P＜0.05). Multivariate analysis indicated that medium-risk CURB-65 score, neutrophil count, LDH, alanine aminotransferase and ≥2 pathogenic microorganisms were independent risk factors for ARF in adult patients with melioidosis pneumonia (odds ratio=21.705, 1.123, 1.010, 1.001, 7.752, all P＜0.05). Predictors screened by Logistic regression were used to construct and validate a nomogram model for predicting ARF in adult patients with melioidosis pneumonia. Receiver operating characteristic curve was plotted to assess model discrimination; the area under the curve was 0.867 (95% confidence interval: 0.700-1.000)(P＜0.05). At the optimal cut-off value defined by the maximum Youden index, the sensitivity was 80.0% and the specificity was 90.0%. Conclusion The model established in this study exhibits favorable predictive performance for ARF complicated by melioidosis pneumonia in adults, and can provide evidence for early clinical warning and timely intervention.